JATSdecoder1.2.2 package

Provides a function collection to extract metadata, sectioned text and study characteristics from scientific articles in 'NISO-JATS' format. Articles in PDF format can be converted to 'NISO-JATS' with the 'Content ExtRactor and MINEr' ('CERMINE', <https://github.com/CeON/CERMINE>). For convenience, two functions bundle the extraction heuristics: JATSdecoder() converts 'NISO-JATS'-tagged XML files to a structured list with elements title, author, journal, history, 'DOI', abstract, sectioned text and reference list. study.character() extracts multiple study characteristics like number of included studies, statistical methods used, alpha error, power, statistical results, correction method for multiple testing, software used. The function get.stats() extracts all statistical results from text and recomputes p-values for many standard test statistics. It performs a consistency check of the reported with the recalculated p-values. An estimation of the involved sample size is performed based on textual reports within the abstract and the reported degrees of freedom within statistical results. In addition, the package contains some useful functions to process text (text2sentences(), text2num(), ngram(), strsplit2(), grep2()). See Böschen, I. (2021) <doi:10.1007/s11192-021-04162-z> Böschen, I. (2021) <doi:10.1038/s41598-021-98782-3>, Böschen, I. (2023) <doi:10.1038/s41598-022-27085-y>, and Böschen, I. (2024) <doi:10.48550/arXiv.2408.07948>.

Useful links

• https://github.com/ingmarboeschen/JATSdecoder/issues
• https://github.com/ingmarboeschen/JATSdecoder